A Shift Towards New Health Information Systems in Liberia

 

Pictured above: sentinel site staff entering data by manually counting malaria cases from the ledger and entering clinical notes directly into laptop. 

In all countries, there is a need for health information.  Health Information Management Systems (HMIS) are struggling all over Africa and Liberia has been given the last chance to see the sentinel site project come to life.  It isn’t easy getting data from a developing country hospital but this has been my work.

I am back in Zwedru and I haven’t blogged in a month.  I was trying to blog daily but my work has put me in front of a computer tinkering with data and making presentations.  Since I last left you, I have had a meeting with USAID on the future of the sentinel sites, switched over to a new Country Director and had an emergency switching of a computer in Zwedru.  Now, I am back with some interesting details.  

Martha Tubman Hospital just came out with the first month of patient level data.  It was a bumpy road as a computer crash left a few days without data.  Also, I am using Dropbox to synchronize the remote hospital folder (with database in it) to my folder in my computer. To those who know Dropbox, it is a file syncing program but I am using it to get my data remotely from a hospital several hundred kilometres away.

For those interested, the staff are no longer tallying, which means that they are no longer counting the number of malaria cases.  Now, all the tallying will be done by the Sentinel Site staff through the paper hospital ledgers and, in parallel, a patient level database, using Epi Info, collects patient information by data entering each clinical note as it goes into the records room.

Now to the interesting part.  THE DATA…

I looked at three methods of collecting health data for the month of March:

1. By entering each record individually into the patient level database
2. By counting the number of malaria cases in the hospital paper ledgers
3. By counting the number of malaria cases which were tallied by the staff themselves.

 

The differences are quite big but here are some preliminary results:

The malaria rate for under 5 years old (number confirmed positive by lab results divided by all those who attended the hospital) was 56.4% for patient level database, 56.1% for hospital ledger and 45.1% for hospital staff tallying.

This means that between 45% to 56% of all hospital visits by children under five were for malaria.

The total malaria death rate (number who died of malaria divided by all those that confirmed positive for malaria in a lab test – also called the malaria case fatality rate) was 0.18%, 0.74% and 0.37% respectively.

This means that less that 1% of those who visited the hospital for malaria died from that disease. 

What is interesting about having patient level data is the ability to make pivot tables in your favourite spreadsheet program and tinker with the values.  Here we see what are the most common secondary diseases if someone already comes in with malaria (separated by age group):

 

Secondary Diagnosis	<5 years old	>=5 years old	Grand Total
Abscess	1	2	3
Amebiasis	8	7	15
Anemia	30	11	41
ARI	180	99	279
Candidasis	16	3	19
Conjunctivitis	2		2
Eye Condition	1	1	2
Hypertension		7	7
Lumbago		2	2
Measles	11	5	16
Otitis	4	1	5
PID		25	25
RIH		2	2
Skin Infection	20	20	40
STD		27	27
Typhoid		8	8
Urine Tract Infection	3	38	41
Worms	18	29	47
Wound	3	9	12
No Secondary Disease	269	500	769
Epilepsy		2	2
Acute Watery Diarrhea	3	4	7
CHAPS		1	1
Meningitis		1	1
Diabetes Mellitus		2	2
Grand Total	569	806	1375

 

As you can see, there were a total of 1375 visits which were diagnosed with malaria and 279 also had ARI (acute respiratory illness) while 27 also had an STD. 

It’s hard to say what you can do with this information right now.  Really, it is more important to get reliable data which is not easy.  Everything from improving the quality of records in the ledgers to renovating the records room is needed.  We’ll see what happens next month…

Pictured above: empty shelves where medicine should be. 

Technorati Tags: Health Information,Liberia,Malaria,Pivot Tables,Health Informatics,Epidemology,EMR,Epi Info,Dropbox

What a day…

I don’t think I have ever felt such frustration and resent as I have felt today.  Right now, I am in Zwedru starting up the new system I have developed for monitoring malaria at this hospital sentinel site.  Emotionally exhausted, I can only reflect on the day and I have realized many things about dealing with authority figures in Liberia.

A few days ago, I was emailed by the Zwedru hospital medical director that he would not accept my choice for a supervisor for the sentinel site.  I held interviews in Monrovia and chose the most suitable candidate.  No suitable candidate came from Zwedru.

I met with the medical director and, at first, I was calm and objective but the fighting began when there became some confusion over who should do the interviewing and where.  Zwedru is far and bringing people there can get expensive.  At the end, the new sentinel site supervisor was accepted but not without complaints to the donors who I still have to answer to when I get back to Monrovia.

After that exhausting adventure, I made my presentation on the new sentinel site system to the data team.  My new system is a shift from the general tallying that occurs in each department, which is an aggregated data describing the activity of the whole hospital, to a patient level data system which shows details of the hospital recorded individually by each patient.  The diagram below describes what I am trying to create here:

As you can see, when the patient enters the hospital, the patient record of chart is merged with the patient and follows them through the patient pathway.  Once the patient leaves the hospital, the hospital record is usually returned to the records room but in my case, the hospital record passes through the data collection officer who documents all the cases that comes through the hospital.  He will document more than just malaria as well, he will record all diseases into an Epi Info Database.  This is a middle ground between a paper based system and a full on electronic medical record (EMR).  This is something I call a “digital medical ledger” (DML) because the output of this database looks more like a ledger than a real database. 

The discussion with the data team was filled with interesting requests for air conditioners and advances training.  Clearly, I cannot give them these things but they ask of course.  I think Liberians see NGOs as being just big bags of money.  It gets frustrating to try to help people who just keep demanding more than they deserve. 

To really figure out how to manage paper data, you have to walk through the hospital and look at how their records move.  For example, in the Outpatients Department (OPD), the hospital record goes from the records room to the clinician carried by a nurse.  The patient gets treated and the record goes to the pharmacy carried by the patient.  The pharmacy dispenses the drugs and holds onto the record until it gets picked up by the nurse, returned to the OPD, documented in a giant ledger and then returned to the records room.  It is fascinating to see all this information flow happening.  I feel a calling in all of this.

So, after all the yelling and screaming, negotiating and refusing, I have come to several conclusions about living here:

1. Don’t offer anything until you actually have it – you will be expected to provide it even if you just said it was a potential possibility.  Also, don’t start a precedent for anything.   The moment you’ve done it once, you will be expected to do it again and again.
2. Make your system perfect the first time.  Liberians don’t change easily.
3. Start your project by dealing with the top most people.  Be firm about what you will offer and don’t promise anything.  Ever.
4. The Liberians will always go back to the main contract to indicate what you have offered.  

I think Liberia is making me jaded.  I have become quick to distrust people and closed to the gifting economy mentality I had back at home.  I understand where I am and how this world works. 

I know I will feel more comfortable once the system I am developing has a consistent rhythm and is working perfectly.  I inherited some problems and I have to go back and clean things up. 

My boss once said “Never lose your sense of humour”.  I hope I am not.  Regardless, it is always in the last place you look….

Technorati Tags: Zwedru,Sentinel Sites,Hospital,EMR

Liberia’s Sentinel Sites for Malaria: An Introduction

 

My work in Liberia has been centred around creating sentinel sites.  A sentinel site is a health facility which collects data on some health indicators and send these indicators to a central location to be reviewed so that improvements to the health care system can be made.  In my case, the goal is to monitor malaria.  Malaria represents a significant portion of outpatient visits   (outpatients are those that come to the hospital to see a doctor and leave the same day – inpatients are those that need to stay IN the hospital overnight.  The majority of patients seen are outpatients.) Across the country, malaria is diagnosed in 44% of all outpatients with the most affected being those under 5 and pregnant women.  For this reason, having dedicated sentinel sites is crucial to the understanding of how malaria, and the programs that are being done to manage the disease, is changing over time. 

To bring it down to the simplest system, a sentinel site just needs to know how many of each subpopulation (<5 years old, over 5 years old, pregnant women, etc…) are being diagnosed with malaria over the total number of patient visits each month.  As a result, we get a percentage of X 5 year old children (numerator) over the entire 5 year old children hospital visits (denominator).  Sounds easy right?  It is just about counting each individual as they come into the hospital and then counting them after they have been diagnosed.  This is called “tallying” since we draw a little line when each person comes in and then count the total number of lines each day and add it up for the whole month.  The problem is that it is not as easy as it sounds.

So far we have two hospitals,  Eternal Love Wins Africa Hospital (ELWA) and Martha Tubman Memorial Hospital (MTM).  ELWA is on the outskirts of Monrovia and MTM is in Zwedru.  The goal is to have six sentinel sites running by the end of next year – one in each area of the country as presented below by the National Malaria Control Program.

The tallying system is done by each department: outpatients department (OPD), inpatient department (IPD), emergency room (ER), antenatal care (ANC) and the obstetrics ward.  Each department has a ledger book which is basically a long lined paper book where the clinical notes are summarized.  Each line is a record of an individual visit and within each line certain features, such as name, age, and final diagnosis (plus many more) are written.  At the end of the day, you count the number of patients you have and the number that have malaria.  Seems simple but think of the enormous amount of extra administrative burden this adds to the clinical staff (mostly nurses).  Lets say each line takes one minute to write down since there are many variable to include in the ledger.  In MTM, there are an average of 200 patients per day passing through the walls of the OPD. This means over three hours of care are lost just doing simple administrative work – work that could be spent dealing with patients.  

So sometimes the clinical staff tally after each patient and sometimes they summarize after the whole day.  Regardless, mistakes are made.  Many mistakes.  Data collected in this way is unaccountable and unreliable but I will get to that in a later post.

My philosophy moves the responsibility of this administrative work from the clinical staff to a data entry clerk.  The basic idea is that a data entry clerk would collect the clinical notes from each department and enter the basic malaria-specific fields into a database.  Then, as we would have patient level data instead of large aggregate data.  The patient level data could be analyzed in many ways – more than through an aggregated tallying system.  There is also a sentinel site supervisor that would double check the data to ensure that each record is entered correctly. 

My database was originally created in Excel but the file became massive and the proprietary nature of the software was presenting limitations (some computers have Excel 2003 and some have Excel 2007).  As a result, I have decided to switch to Epi Info and buit the database there and do all the analysis elsewhere – most likely Excel.

What am I measuring?  Here is the list and don’t worry if you don’t understand all the terms but please ask if you are interested:

Outpatient Cases

• Number of total outpatients (<5, 5+)
• Number of outpatient suspect malaria cases (<5, 5+)
• Number of outpatients lab tested for malaria with slides and/or RDTs (<5, 5+)
• Number of outpatient lab-confirmed malaria cases with slides and/or RDTs (<5, 5+)

Inpatient Cases

• Number of total inpatients (<5, 5+)
• Number of inpatient suspect malaria cases (<5, 5+)
• Number of inpatients lab tested for malaria with slides and/or RDTs (<5, 5+)
• Number of inpatient lab-confirmed malaria cases with slides and/or RDTs (<5, 5+)
• Number of inpatient malaria deaths (<5, 5+)
• Number of inpatient anemia cases (<5)

Treatment

• Number of antimalarial treatments prescribed, by type of Tx (<5, 5+)
• Number of days out of stock of commodities, by type of commodity (ACT, slides, RDTs)
• Number of children <5 receiving a blood transfusion

IPTp

• Number of pregnant women attending first ANC visit
• Number of pregnant women who received IPTp-1
• Number of pregnant women who received IPTp-2

As simple as the philosophy sounds, it is fraught with issues.  I have to understand how information moves through a hospital.  What is the volume of patients come through the hospital?  Do the hospitals have a records room and if so, how many (some have an extra one for antenatal care)?  How many beds in the inpatients?  At what times does the hospital have power?  For Zwedru, the power comes on from 10am – 3pm then after 6pm.  There are also fine social dynamics that are played out between the staff and MENTOR as we try to collaborate on getting this data but that will be detailed later as well.

Any questions?

Oh, and for those interested to know what the picture is at the top of the post – it is the hospital records room at the Buchanan Hospital.  Can you imagine trying to find a record there?

Technorati Tags: sentinel sites,hospital records,EMR,malaria